icu_properties/script.rs
1// This file is part of ICU4X. For terms of use, please see the file
2// called LICENSE at the top level of the ICU4X source tree
3// (online at: https://github.com/unicode-org/icu4x/blob/main/LICENSE ).
4
5//! Data and APIs for supporting Script_Extensions property
6//! values in an efficient structure.
7
8use crate::props::Script;
9use crate::provider::*;
10
11#[cfg(feature = "alloc")]
12use core::iter::FromIterator;
13use core::ops::RangeInclusive;
14#[cfg(feature = "alloc")]
15use icu_collections::codepointinvlist::CodePointInversionList;
16use icu_provider::prelude::*;
17use zerovec::{ule::AsULE, ZeroSlice};
18
19/// The number of bits at the low-end of a `ScriptWithExt` value used for
20/// storing the `Script` value (or `extensions` index).
21const SCRIPT_VAL_LENGTH: u16 = 10;
22
23/// The bit mask necessary to retrieve the `Script` value (or `extensions` index)
24/// from a `ScriptWithExt` value.
25const SCRIPT_X_SCRIPT_VAL: u16 = (1 << SCRIPT_VAL_LENGTH) - 1;
26
27/// An internal-use only pseudo-property that represents the values stored in
28/// the trie of the special data structure [`ScriptWithExtensionsProperty`].
29///
30/// Note: The will assume a 12-bit layout. The 2 higher order bits in positions
31/// 11..10 will indicate how to deduce the Script value and Script_Extensions,
32/// and the lower 10 bits 9..0 indicate either the Script value or the index
33/// into the `extensions` structure.
34#[derive(Copy, Clone, Debug, Eq, PartialEq)]
35#[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
36#[cfg_attr(feature = "datagen", derive(databake::Bake))]
37#[cfg_attr(feature = "datagen", databake(path = icu_properties::script))]
38#[repr(transparent)]
39#[doc(hidden)]
40// `ScriptWithExt` not intended as public-facing but for `ScriptWithExtensionsProperty` constructor
41#[allow(clippy::exhaustive_structs)] // this type is stable
42pub struct ScriptWithExt(pub u16);
43
44#[allow(missing_docs)] // These constants don't need individual documentation.
45#[allow(non_upper_case_globals)]
46#[doc(hidden)] // `ScriptWithExt` not intended as public-facing but for `ScriptWithExtensionsProperty` constructor
47impl ScriptWithExt {
48 pub const Unknown: ScriptWithExt = ScriptWithExt(0);
49}
50
51impl AsULE for ScriptWithExt {
52 type ULE = <u16 as AsULE>::ULE;
53
54 #[inline]
55 fn to_unaligned(self) -> Self::ULE {
56 Script(self.0).to_unaligned()
57 }
58
59 #[inline]
60 fn from_unaligned(unaligned: Self::ULE) -> Self {
61 ScriptWithExt(Script::from_unaligned(unaligned).0)
62 }
63}
64
65#[doc(hidden)] // `ScriptWithExt` not intended as public-facing but for `ScriptWithExtensionsProperty` constructor
66impl ScriptWithExt {
67 /// Returns whether the [`ScriptWithExt`] value has Script_Extensions and
68 /// also indicates a Script value of [`Script::Common`].
69 ///
70 /// # Examples
71 ///
72 /// ```
73 /// use icu::properties::script::ScriptWithExt;
74 ///
75 /// assert!(ScriptWithExt(0x04FF).is_common());
76 /// assert!(ScriptWithExt(0x0400).is_common());
77 ///
78 /// assert!(!ScriptWithExt(0x08FF).is_common());
79 /// assert!(!ScriptWithExt(0x0800).is_common());
80 ///
81 /// assert!(!ScriptWithExt(0x0CFF).is_common());
82 /// assert!(!ScriptWithExt(0x0C00).is_common());
83 ///
84 /// assert!(!ScriptWithExt(0xFF).is_common());
85 /// assert!(!ScriptWithExt(0x0).is_common());
86 /// ```
87 pub fn is_common(&self) -> bool {
88 self.0 >> SCRIPT_VAL_LENGTH == 1
89 }
90
91 /// Returns whether the [`ScriptWithExt`] value has Script_Extensions and
92 /// also indicates a Script value of [`Script::Inherited`].
93 ///
94 /// # Examples
95 ///
96 /// ```
97 /// use icu::properties::script::ScriptWithExt;
98 ///
99 /// assert!(!ScriptWithExt(0x04FF).is_inherited());
100 /// assert!(!ScriptWithExt(0x0400).is_inherited());
101 ///
102 /// assert!(ScriptWithExt(0x08FF).is_inherited());
103 /// assert!(ScriptWithExt(0x0800).is_inherited());
104 ///
105 /// assert!(!ScriptWithExt(0x0CFF).is_inherited());
106 /// assert!(!ScriptWithExt(0x0C00).is_inherited());
107 ///
108 /// assert!(!ScriptWithExt(0xFF).is_inherited());
109 /// assert!(!ScriptWithExt(0x0).is_inherited());
110 /// ```
111 pub fn is_inherited(&self) -> bool {
112 self.0 >> SCRIPT_VAL_LENGTH == 2
113 }
114
115 /// Returns whether the [`ScriptWithExt`] value has Script_Extensions and
116 /// also indicates that the Script value is neither [`Script::Common`] nor
117 /// [`Script::Inherited`].
118 ///
119 /// # Examples
120 ///
121 /// ```
122 /// use icu::properties::script::ScriptWithExt;
123 ///
124 /// assert!(!ScriptWithExt(0x04FF).is_other());
125 /// assert!(!ScriptWithExt(0x0400).is_other());
126 ///
127 /// assert!(!ScriptWithExt(0x08FF).is_other());
128 /// assert!(!ScriptWithExt(0x0800).is_other());
129 ///
130 /// assert!(ScriptWithExt(0x0CFF).is_other());
131 /// assert!(ScriptWithExt(0x0C00).is_other());
132 ///
133 /// assert!(!ScriptWithExt(0xFF).is_other());
134 /// assert!(!ScriptWithExt(0x0).is_other());
135 /// ```
136 pub fn is_other(&self) -> bool {
137 self.0 >> SCRIPT_VAL_LENGTH == 3
138 }
139
140 /// Returns whether the [`ScriptWithExt`] value has Script_Extensions.
141 ///
142 /// # Examples
143 ///
144 /// ```
145 /// use icu::properties::script::ScriptWithExt;
146 ///
147 /// assert!(ScriptWithExt(0x04FF).has_extensions());
148 /// assert!(ScriptWithExt(0x0400).has_extensions());
149 ///
150 /// assert!(ScriptWithExt(0x08FF).has_extensions());
151 /// assert!(ScriptWithExt(0x0800).has_extensions());
152 ///
153 /// assert!(ScriptWithExt(0x0CFF).has_extensions());
154 /// assert!(ScriptWithExt(0x0C00).has_extensions());
155 ///
156 /// assert!(!ScriptWithExt(0xFF).has_extensions());
157 /// assert!(!ScriptWithExt(0x0).has_extensions());
158 /// ```
159 pub fn has_extensions(&self) -> bool {
160 let high_order_bits = self.0 >> SCRIPT_VAL_LENGTH;
161 high_order_bits > 0
162 }
163}
164
165impl From<ScriptWithExt> for u32 {
166 fn from(swe: ScriptWithExt) -> Self {
167 swe.0 as u32
168 }
169}
170
171impl From<ScriptWithExt> for Script {
172 fn from(swe: ScriptWithExt) -> Self {
173 Script(swe.0)
174 }
175}
176
177/// A struct that wraps a [`Script`] array, such as in the return value for
178/// [`get_script_extensions_val()`](ScriptWithExtensionsBorrowed::get_script_extensions_val).
179#[derive(Copy, Clone, Debug, Eq, PartialEq)]
180pub struct ScriptExtensionsSet<'a> {
181 values: &'a ZeroSlice<Script>,
182}
183
184impl<'a> ScriptExtensionsSet<'a> {
185 /// Returns whether this set contains the given script.
186 ///
187 /// # Example
188 ///
189 /// ```
190 /// use icu::properties::props::Script;
191 /// use icu::properties::script::ScriptWithExtensions;
192 /// let swe = ScriptWithExtensions::new();
193 ///
194 /// assert!(swe
195 /// .get_script_extensions_val('\u{11303}') // GRANTHA SIGN VISARGA
196 /// .contains(&Script::Grantha));
197 /// ```
198 pub fn contains(&self, x: &Script) -> bool {
199 ZeroSlice::binary_search(self.values, x).is_ok()
200 }
201
202 /// Gets an iterator over the elements.
203 ///
204 /// # Example
205 ///
206 /// ```
207 /// use icu::properties::props::Script;
208 /// use icu::properties::script::ScriptWithExtensions;
209 /// let swe = ScriptWithExtensions::new();
210 ///
211 /// assert_eq!(
212 /// swe.get_script_extensions_val('௫') // U+0BEB TAMIL DIGIT FIVE
213 /// .iter()
214 /// .collect::<Vec<_>>(),
215 /// [Script::Tamil, Script::Grantha]
216 /// );
217 /// ```
218 pub fn iter(&self) -> impl DoubleEndedIterator<Item = Script> + 'a {
219 ZeroSlice::iter(self.values)
220 }
221
222 /// For accessing this set as an array instead of an iterator
223 #[doc(hidden)] // used by FFI code
224 pub fn array_len(&self) -> usize {
225 self.values.len()
226 }
227 /// For accessing this set as an array instead of an iterator
228 #[doc(hidden)] // used by FFI code
229 pub fn array_get(&self, index: usize) -> Option<Script> {
230 self.values.get(index)
231 }
232}
233
234/// A struct that represents the data for the Script and Script_Extensions properties.
235///
236/// ✨ *Enabled with the `compiled_data` Cargo feature.*
237///
238/// [📚 Help choosing a constructor](icu_provider::constructors)
239///
240/// Most useful methods are on [`ScriptWithExtensionsBorrowed`] obtained by calling [`ScriptWithExtensions::as_borrowed()`]
241///
242/// # Examples
243///
244/// ```
245/// use icu::properties::script::ScriptWithExtensions;
246/// use icu::properties::props::Script;
247/// let swe = ScriptWithExtensions::new();
248///
249/// // get the `Script` property value
250/// assert_eq!(swe.get_script_val('ـ'), Script::Common); // U+0640 ARABIC TATWEEL
251/// assert_eq!(swe.get_script_val('\u{0650}'), Script::Inherited); // U+0650 ARABIC KASRA
252/// assert_eq!(swe.get_script_val('٠'), Script::Arabic); // // U+0660 ARABIC-INDIC DIGIT ZERO
253/// assert_eq!(swe.get_script_val('ﷲ'), Script::Arabic); // U+FDF2 ARABIC LIGATURE ALLAH ISOLATED FORM
254///
255/// // get the `Script_Extensions` property value
256/// assert_eq!(
257/// swe.get_script_extensions_val('ـ') // U+0640 ARABIC TATWEEL
258/// .iter().collect::<Vec<_>>(),
259/// [Script::Arabic, Script::Syriac, Script::Mandaic, Script::Manichaean,
260/// Script::PsalterPahlavi, Script::Adlam, Script::HanifiRohingya, Script::Sogdian,
261/// Script::OldUyghur]
262/// );
263/// assert_eq!(
264/// swe.get_script_extensions_val('🥳') // U+1F973 FACE WITH PARTY HORN AND PARTY HAT
265/// .iter().collect::<Vec<_>>(),
266/// [Script::Common]
267/// );
268/// assert_eq!(
269/// swe.get_script_extensions_val('\u{200D}') // ZERO WIDTH JOINER
270/// .iter().collect::<Vec<_>>(),
271/// [Script::Inherited]
272/// );
273/// assert_eq!(
274/// swe.get_script_extensions_val('௫') // U+0BEB TAMIL DIGIT FIVE
275/// .iter().collect::<Vec<_>>(),
276/// [Script::Tamil, Script::Grantha]
277/// );
278///
279/// // check containment of a `Script` value in the `Script_Extensions` value
280/// // U+0650 ARABIC KASRA
281/// assert!(!swe.has_script('\u{0650}', Script::Inherited)); // main Script value
282/// assert!(swe.has_script('\u{0650}', Script::Arabic));
283/// assert!(swe.has_script('\u{0650}', Script::Syriac));
284/// assert!(!swe.has_script('\u{0650}', Script::Thaana));
285///
286/// // get a `CodePointInversionList` for when `Script` value is contained in `Script_Extensions` value
287/// let syriac = swe.get_script_extensions_set(Script::Syriac);
288/// assert!(syriac.contains('\u{0650}')); // ARABIC KASRA
289/// assert!(!syriac.contains('٠')); // ARABIC-INDIC DIGIT ZERO
290/// assert!(!syriac.contains('ﷲ')); // ARABIC LIGATURE ALLAH ISOLATED FORM
291/// assert!(syriac.contains('܀')); // SYRIAC END OF PARAGRAPH
292/// assert!(syriac.contains('\u{074A}')); // SYRIAC BARREKH
293/// ```
294#[derive(Debug)]
295pub struct ScriptWithExtensions {
296 data: DataPayload<PropertyScriptWithExtensionsV1>,
297}
298
299/// A borrowed wrapper around script extension data, returned by
300/// [`ScriptWithExtensions::as_borrowed()`]. More efficient to query.
301#[derive(Clone, Copy, Debug)]
302pub struct ScriptWithExtensionsBorrowed<'a> {
303 data: &'a ScriptWithExtensionsProperty<'a>,
304}
305
306impl ScriptWithExtensions {
307 /// Creates a new instance of `ScriptWithExtensionsBorrowed` using compiled data.
308 ///
309 /// ✨ *Enabled with the `compiled_data` Cargo feature.*
310 ///
311 /// [📚 Help choosing a constructor](icu_provider::constructors)
312 #[cfg(feature = "compiled_data")]
313 #[allow(clippy::new_ret_no_self)]
314 pub fn new() -> ScriptWithExtensionsBorrowed<'static> {
315 ScriptWithExtensionsBorrowed::new()
316 }
317
318 icu_provider::gen_buffer_data_constructors!(
319 () -> result: Result<ScriptWithExtensions, DataError>,
320 functions: [
321 new: skip,
322 try_new_with_buffer_provider,
323 try_new_unstable,
324 Self,
325 ]
326 );
327
328 #[doc = icu_provider::gen_buffer_unstable_docs!(UNSTABLE, Self::new)]
329 pub fn try_new_unstable(
330 provider: &(impl DataProvider<PropertyScriptWithExtensionsV1> + ?Sized),
331 ) -> Result<Self, DataError> {
332 Ok(ScriptWithExtensions::from_data(
333 provider.load(Default::default())?.payload,
334 ))
335 }
336
337 /// Construct a borrowed version of this type that can be queried.
338 ///
339 /// This avoids a potential small underlying cost per API call (ex: `contains()`) by consolidating it
340 /// up front.
341 #[inline]
342 pub fn as_borrowed(&self) -> ScriptWithExtensionsBorrowed<'_> {
343 ScriptWithExtensionsBorrowed {
344 data: self.data.get(),
345 }
346 }
347
348 /// Construct a new one from loaded data
349 ///
350 /// Typically it is preferable to use getters like [`load_script_with_extensions_unstable()`] instead
351 pub(crate) fn from_data(data: DataPayload<PropertyScriptWithExtensionsV1>) -> Self {
352 Self { data }
353 }
354}
355
356impl<'a> ScriptWithExtensionsBorrowed<'a> {
357 /// Returns the `Script` property value for this code point.
358 ///
359 /// # Examples
360 ///
361 /// ```
362 /// use icu::properties::script::ScriptWithExtensions;
363 /// use icu::properties::props::Script;
364 ///
365 /// let swe = ScriptWithExtensions::new();
366 ///
367 /// // U+0640 ARABIC TATWEEL
368 /// assert_eq!(swe.get_script_val('ـ'), Script::Common); // main Script value
369 /// assert_ne!(swe.get_script_val('ـ'), Script::Arabic);
370 /// assert_ne!(swe.get_script_val('ـ'), Script::Syriac);
371 /// assert_ne!(swe.get_script_val('ـ'), Script::Thaana);
372 ///
373 /// // U+0650 ARABIC KASRA
374 /// assert_eq!(swe.get_script_val('\u{0650}'), Script::Inherited); // main Script value
375 /// assert_ne!(swe.get_script_val('\u{0650}'), Script::Arabic);
376 /// assert_ne!(swe.get_script_val('\u{0650}'), Script::Syriac);
377 /// assert_ne!(swe.get_script_val('\u{0650}'), Script::Thaana);
378 ///
379 /// // U+0660 ARABIC-INDIC DIGIT ZERO
380 /// assert_ne!(swe.get_script_val('٠'), Script::Common);
381 /// assert_eq!(swe.get_script_val('٠'), Script::Arabic); // main Script value
382 /// assert_ne!(swe.get_script_val('٠'), Script::Syriac);
383 /// assert_ne!(swe.get_script_val('٠'), Script::Thaana);
384 ///
385 /// // U+FDF2 ARABIC LIGATURE ALLAH ISOLATED FORM
386 /// assert_ne!(swe.get_script_val('ﷲ'), Script::Common);
387 /// assert_eq!(swe.get_script_val('ﷲ'), Script::Arabic); // main Script value
388 /// assert_ne!(swe.get_script_val('ﷲ'), Script::Syriac);
389 /// assert_ne!(swe.get_script_val('ﷲ'), Script::Thaana);
390 /// ```
391 pub fn get_script_val(self, ch: char) -> Script {
392 self.get_script_val32(ch as u32)
393 }
394
395 /// See [`Self::get_script_val`].
396 pub fn get_script_val32(self, code_point: u32) -> Script {
397 let sc_with_ext = self.data.trie.get32(code_point);
398
399 if sc_with_ext.is_other() {
400 let ext_idx = sc_with_ext.0 & SCRIPT_X_SCRIPT_VAL;
401 let scx_val = self.data.extensions.get(ext_idx as usize);
402 let scx_first_sc = scx_val.and_then(|scx| scx.get(0));
403
404 let default_sc_val = Script::Unknown;
405
406 scx_first_sc.unwrap_or(default_sc_val)
407 } else if sc_with_ext.is_common() {
408 Script::Common
409 } else if sc_with_ext.is_inherited() {
410 Script::Inherited
411 } else {
412 let script_val = sc_with_ext.0;
413 Script(script_val)
414 }
415 }
416 // Returns the Script_Extensions value for a code_point when the trie value
417 // is already known.
418 // This private helper method exists to prevent code duplication in callers like
419 // `get_script_extensions_val`, `get_script_extensions_set`, and `has_script`.
420 fn get_scx_val_using_trie_val(
421 self,
422 sc_with_ext_ule: &'a <ScriptWithExt as AsULE>::ULE,
423 ) -> &'a ZeroSlice<Script> {
424 let sc_with_ext = ScriptWithExt::from_unaligned(*sc_with_ext_ule);
425 if sc_with_ext.is_other() {
426 let ext_idx = sc_with_ext.0 & SCRIPT_X_SCRIPT_VAL;
427 let ext_subarray = self.data.extensions.get(ext_idx as usize);
428 // In the OTHER case, where the 2 higher-order bits of the
429 // `ScriptWithExt` value in the trie doesn't indicate the Script value,
430 // the Script value is copied/inserted into the first position of the
431 // `extensions` array. So we must remove it to return the actual scx array val.
432 let scx_slice = ext_subarray
433 .and_then(|zslice| zslice.as_ule_slice().get(1..))
434 .unwrap_or_default();
435 ZeroSlice::from_ule_slice(scx_slice)
436 } else if sc_with_ext.is_common() || sc_with_ext.is_inherited() {
437 let ext_idx = sc_with_ext.0 & SCRIPT_X_SCRIPT_VAL;
438 let scx_val = self.data.extensions.get(ext_idx as usize);
439 scx_val.unwrap_or_default()
440 } else {
441 // Note: `Script` and `ScriptWithExt` are both represented as the same
442 // u16 value when the `ScriptWithExt` has no higher-order bits set.
443 let script_ule_slice = core::slice::from_ref(sc_with_ext_ule);
444 ZeroSlice::from_ule_slice(script_ule_slice)
445 }
446 }
447 /// Return the `Script_Extensions` property value for this code point.
448 ///
449 /// If `code_point` has Script_Extensions, then return the Script codes in
450 /// the Script_Extensions. In this case, the Script property value
451 /// (normally Common or Inherited) is not included in the [`ScriptExtensionsSet`].
452 ///
453 /// If c does not have Script_Extensions, then the one Script code is put
454 /// into the [`ScriptExtensionsSet`] and also returned.
455 ///
456 /// If c is not a valid code point, then return an empty [`ScriptExtensionsSet`].
457 ///
458 /// # Examples
459 ///
460 /// ```
461 /// use icu::properties::script::ScriptWithExtensions;
462 /// use icu::properties::props::Script;
463 ///
464 /// let swe = ScriptWithExtensions::new();
465 ///
466 /// assert_eq!(
467 /// swe.get_script_extensions_val('𐓐') // U+104D0 OSAGE CAPITAL LETTER KHA
468 /// .iter()
469 /// .collect::<Vec<_>>(),
470 /// [Script::Osage]
471 /// );
472 /// assert_eq!(
473 /// swe.get_script_extensions_val('🥳') // U+1F973 FACE WITH PARTY HORN AND PARTY HAT
474 /// .iter()
475 /// .collect::<Vec<_>>(),
476 /// [Script::Common]
477 /// );
478 /// assert_eq!(
479 /// swe.get_script_extensions_val('\u{200D}') // ZERO WIDTH JOINER
480 /// .iter()
481 /// .collect::<Vec<_>>(),
482 /// [Script::Inherited]
483 /// );
484 /// assert_eq!(
485 /// swe.get_script_extensions_val('௫') // U+0BEB TAMIL DIGIT FIVE
486 /// .iter()
487 /// .collect::<Vec<_>>(),
488 /// [Script::Tamil, Script::Grantha]
489 /// );
490 /// ```
491 pub fn get_script_extensions_val(self, ch: char) -> ScriptExtensionsSet<'a> {
492 self.get_script_extensions_val32(ch as u32)
493 }
494
495 /// See [`Self::get_script_extensions_val`].
496 pub fn get_script_extensions_val32(self, code_point: u32) -> ScriptExtensionsSet<'a> {
497 let sc_with_ext_ule = self.data.trie.get32_ule(code_point);
498
499 ScriptExtensionsSet {
500 values: match sc_with_ext_ule {
501 Some(ule_ref) => self.get_scx_val_using_trie_val(ule_ref),
502 None => ZeroSlice::from_ule_slice(&[]),
503 },
504 }
505 }
506
507 /// Returns whether `script` is contained in the Script_Extensions
508 /// property value if the code_point has Script_Extensions, otherwise
509 /// if the code point does not have Script_Extensions then returns
510 /// whether the Script property value matches.
511 ///
512 /// Some characters are commonly used in multiple scripts. For more information,
513 /// see UAX #24: <http://www.unicode.org/reports/tr24/>.
514 ///
515 /// # Examples
516 ///
517 /// ```
518 /// use icu::properties::script::ScriptWithExtensions;
519 /// use icu::properties::props::Script;
520 ///
521 /// let swe = ScriptWithExtensions::new();
522 ///
523 /// // U+0650 ARABIC KASRA
524 /// assert!(!swe.has_script('\u{0650}', Script::Inherited)); // main Script value
525 /// assert!(swe.has_script('\u{0650}', Script::Arabic));
526 /// assert!(swe.has_script('\u{0650}', Script::Syriac));
527 /// assert!(!swe.has_script('\u{0650}', Script::Thaana));
528 ///
529 /// // U+0660 ARABIC-INDIC DIGIT ZERO
530 /// assert!(!swe.has_script('٠', Script::Common)); // main Script value
531 /// assert!(swe.has_script('٠', Script::Arabic));
532 /// assert!(!swe.has_script('٠', Script::Syriac));
533 /// assert!(swe.has_script('٠', Script::Thaana));
534 ///
535 /// // U+FDF2 ARABIC LIGATURE ALLAH ISOLATED FORM
536 /// assert!(!swe.has_script('ﷲ', Script::Common));
537 /// assert!(swe.has_script('ﷲ', Script::Arabic)); // main Script value
538 /// assert!(!swe.has_script('ﷲ', Script::Syriac));
539 /// assert!(swe.has_script('ﷲ', Script::Thaana));
540 /// ```
541 pub fn has_script(self, ch: char, script: Script) -> bool {
542 self.has_script32(ch as u32, script)
543 }
544
545 /// See [`Self::has_script`].
546 pub fn has_script32(self, code_point: u32, script: Script) -> bool {
547 let sc_with_ext_ule = if let Some(scwe_ule) = self.data.trie.get32_ule(code_point) {
548 scwe_ule
549 } else {
550 return false;
551 };
552 let sc_with_ext = <ScriptWithExt as AsULE>::from_unaligned(*sc_with_ext_ule);
553
554 if !sc_with_ext.has_extensions() {
555 let script_val = sc_with_ext.0;
556 script == Script(script_val)
557 } else {
558 let scx_val = self.get_scx_val_using_trie_val(sc_with_ext_ule);
559 let script_find = scx_val.iter().find(|&sc| sc == script);
560 script_find.is_some()
561 }
562 }
563
564 /// Returns all of the matching `CodePointMapRange`s for the given [`Script`]
565 /// in which `has_script` will return true for all of the contained code points.
566 ///
567 /// # Examples
568 ///
569 /// ```
570 /// use icu::properties::props::Script;
571 /// use icu::properties::script::ScriptWithExtensions;
572 ///
573 /// let swe = ScriptWithExtensions::new();
574 ///
575 /// let syriac_script_extensions_ranges =
576 /// swe.get_script_extensions_ranges(Script::Syriac);
577 ///
578 /// let exp_ranges = [
579 /// 0x0303..=0x0304, // COMBINING TILDE..COMBINING MACRON
580 /// 0x0307..=0x0308, // COMBINING DOT ABOVE..COMBINING DIAERESIS
581 /// 0x030A..=0x030A, // COMBINING RING ABOVE
582 /// 0x0320..=0x0320, // COMBINING MINUS SIGN BELOW
583 /// 0x0323..=0x0325, // COMBINING DOT BELOW..COMBINING RING BELOW
584 /// 0x032D..=0x032E, // COMBINING CIRCUMFLEX ACCENT BELOW..COMBINING BREVE BELOW
585 /// 0x0330..=0x0330, // COMBINING TILDE BELOW
586 /// 0x060C..=0x060C, // ARABIC COMMA
587 /// 0x061B..=0x061C, // ARABIC SEMICOLON, ARABIC LETTER MARK
588 /// 0x061F..=0x061F, // ARABIC QUESTION MARK
589 /// 0x0640..=0x0640, // ARABIC TATWEEL
590 /// 0x064B..=0x0655, // ARABIC FATHATAN..ARABIC HAMZA BELOW
591 /// 0x0670..=0x0670, // ARABIC LETTER SUPERSCRIPT ALEF
592 /// 0x0700..=0x070D, // Syriac block begins at U+0700
593 /// 0x070F..=0x074A, // Syriac block
594 /// 0x074D..=0x074F, // Syriac block ends at U+074F
595 /// 0x0860..=0x086A, // Syriac Supplement block is U+0860..=U+086F
596 /// 0x1DF8..=0x1DF8, // COMBINING DOT ABOVE LEFT
597 /// 0x1DFA..=0x1DFA, // COMBINING DOT BELOW LEFT
598 /// ];
599 ///
600 /// assert_eq!(
601 /// syriac_script_extensions_ranges.collect::<Vec<_>>(),
602 /// exp_ranges
603 /// );
604 /// ```
605 pub fn get_script_extensions_ranges(
606 self,
607 script: Script,
608 ) -> impl Iterator<Item = RangeInclusive<u32>> + 'a {
609 self.data
610 .trie
611 .iter_ranges_mapped(move |value| {
612 let sc_with_ext = ScriptWithExt(value.0);
613 if sc_with_ext.has_extensions() {
614 self.get_scx_val_using_trie_val(&sc_with_ext.to_unaligned())
615 .iter()
616 .any(|sc| sc == script)
617 } else {
618 script == sc_with_ext.into()
619 }
620 })
621 .filter(|v| v.value)
622 .map(|v| v.range)
623 }
624
625 /// Returns a [`CodePointInversionList`] for the given [`Script`] which represents all
626 /// code points for which `has_script` will return true.
627 ///
628 /// # Examples
629 ///
630 /// ```
631 /// use icu::properties::script::ScriptWithExtensions;
632 /// use icu::properties::props::Script;
633 ///
634 /// let swe = ScriptWithExtensions::new();
635 ///
636 /// let syriac = swe.get_script_extensions_set(Script::Syriac);
637 ///
638 /// assert!(!syriac.contains('؞')); // ARABIC TRIPLE DOT PUNCTUATION MARK
639 /// assert!(syriac.contains('؟')); // ARABIC QUESTION MARK
640 /// assert!(!syriac.contains('ؠ')); // ARABIC LETTER KASHMIRI YEH
641 ///
642 /// assert!(syriac.contains('܀')); // SYRIAC END OF PARAGRAPH
643 /// assert!(syriac.contains('\u{074A}')); // SYRIAC BARREKH
644 /// assert!(!syriac.contains('\u{074B}')); // unassigned
645 /// assert!(syriac.contains('ݏ')); // SYRIAC LETTER SOGDIAN FE
646 /// assert!(!syriac.contains('ݐ')); // ARABIC LETTER BEH WITH THREE DOTS HORIZONTALLY BELOW
647 ///
648 /// assert!(syriac.contains('\u{1DF8}')); // COMBINING DOT ABOVE LEFT
649 /// assert!(!syriac.contains('\u{1DF9}')); // COMBINING WIDE INVERTED BRIDGE BELOW
650 /// assert!(syriac.contains('\u{1DFA}')); // COMBINING DOT BELOW LEFT
651 /// assert!(!syriac.contains('\u{1DFB}')); // COMBINING DELETION MARK
652 /// ```
653 #[cfg(feature = "alloc")]
654 pub fn get_script_extensions_set(self, script: Script) -> CodePointInversionList<'a> {
655 CodePointInversionList::from_iter(self.get_script_extensions_ranges(script))
656 }
657}
658
659#[cfg(feature = "compiled_data")]
660impl Default for ScriptWithExtensionsBorrowed<'static> {
661 fn default() -> Self {
662 Self::new()
663 }
664}
665
666impl ScriptWithExtensionsBorrowed<'static> {
667 /// Creates a new instance of `ScriptWithExtensionsBorrowed` using compiled data.
668 ///
669 /// ✨ *Enabled with the `compiled_data` Cargo feature.*
670 ///
671 /// [📚 Help choosing a constructor](icu_provider::constructors)
672 #[cfg(feature = "compiled_data")]
673 pub fn new() -> Self {
674 Self {
675 data: crate::provider::Baked::SINGLETON_PROPERTY_SCRIPT_WITH_EXTENSIONS_V1,
676 }
677 }
678
679 /// Cheaply converts a [`ScriptWithExtensionsBorrowed<'static>`] into a [`ScriptWithExtensions`].
680 ///
681 /// Note: Due to branching and indirection, using [`ScriptWithExtensions`] might inhibit some
682 /// compile-time optimizations that are possible with [`ScriptWithExtensionsBorrowed`].
683 pub const fn static_to_owned(self) -> ScriptWithExtensions {
684 ScriptWithExtensions {
685 data: DataPayload::from_static_ref(self.data),
686 }
687 }
688}
689
690#[cfg(test)]
691mod tests {
692 use super::*;
693 #[test]
694 /// Regression test for https://github.com/unicode-org/icu4x/issues/6041
695 fn test_scx_regression_6041() {
696 let scripts = ScriptWithExtensions::new()
697 .get_script_extensions_val('\u{2bc}')
698 .iter()
699 .collect::<Vec<_>>();
700 assert_eq!(
701 scripts,
702 [
703 Script::Bengali,
704 Script::Cyrillic,
705 Script::Devanagari,
706 Script::Latin,
707 Script::Thai,
708 Script::Lisu,
709 Script::Toto
710 ]
711 );
712 }
713}